Conservation threats and the phylogenetic utility of IUCN Red List rankings in Incilius toads
Sandra R. Schachat,
∗† Daniel G. Mulcahy,‡ and Joseph R. Mendelson III§
∗∗¶
∗Department of Biochemistry, Molecular Biology, Entomology & Plant Pathology, Mississippi State University, Mississippi State, MS, 39762 U.S.A.
†Department of Paleobiology, National Museum of Natural History, Smithsonian Institution, P.O. Box 37012, MRC 121, Washington, D.C. 20013, U.S.A.
‡Global Genome Initiative, National Museum of Natural History, Smithsonian Institution, P.O. Box 37012, MRC 183, Washington, D.C. 20013, U.S.A.
§Zoo Atlanta, 800 Cherokee Ave SE, Atlanta, GA 30315, U.S.A.
∗∗School of Biology, Georgia Institute of Technology, 301 Ferst Drive, Atlanta, GA 30332, U.S.A.
Abstract: Phylogenetic analysis of extinction threat is an emerging tool in the field of conservation. However, there are problems with the methods and data as commonly used. Phylogenetic sampling usually extends to the level of family or genus, but International Union for Conservation of Nature (IUCN) rankings are available only for individual species, and, although different species within a taxonomic group may have the same IUCN rank, the species may have been ranked as such for different reasons. Therefore, IUCN rank may not reflect evolutionary history and thus may not be appropriate for use in a phylogenetic context. To be used appropriately, threat-risk data should reflect the cause of extinction threat rather than the IUCN threat ranking. In a case study of the toad genusIncilius, with phylogenetic sampling at the species level (so that the resolution of the phylogeny matches character data from the IUCN Red List), we analyzed causes of decline and IUCN threat rankings by calculating metrics of phylogenetic signal (such as Fritz and Purvis’ D). We also analyzed the extent to which cause of decline and threat ranking overlap by calculating phylogenetic correlation between these 2 types of character data. Inciliusspecies varied greatly in both threat ranking and cause of decline; this variability would be lost at a coarser taxonomic resolution. We found far more phylogenetic signal, likely correlated with evolutionary history, for causes of decline than for IUCN threat ranking. Individual causes of decline and IUCN threat rankings were largely uncorrelated on the phylogeny.
Our results demonstrate the importance of character selection and taxonomic resolution when extinction threat is analyzed in a phylogenetic context.
Keywords: amphibian, Bufonidae, extinction threat, Mesoamerica, phylogeny
Las Amenazas para la Conservaci´on y la Utilidad Filogen´etica de las Clasificaciones de la Lista Roja de la UICN de los SaposIncilius
Resumen: El an´alisis filogen´etico de la amenaza de extinci´on es una herramienta emergente en el campo de la conservaci´on. Sin embargo, existen problemas con los m´etodos y los datos que se usan com´unmente. El muestreo filogen´etico generalmente se extiende hasta el nivel de familia o g´enero, pero las clasificaciones de la Uni´on Internacional para la Conservaci´on de la Naturaleza (UICN) est´an disponibles solamente para especies individuales, y aunque especies diferentes puedan tener la misma clasificaci´on de la UICN, la especie puede estar clasificada as´ı por razones distintas. Por eso, la clasificaci´on de la UICN puede no reflejar la historia evolutiva y por lo tanto no ser la adecuada en un contexto filogen´etico. Para que se use adecuadamente, la informaci´on de amenaza de riesgo debe reflejar la causa de la amenaza de extinci´on en lugar de la clasificaci´on de amenaza de la UICN. En un estudio de caso del g´enero de sapos Incilius, con muestreo filogen´etico a nivel de especies (para que la resoluci´on de la filogenia sea igual a datos de caracter´ısticas de la Lista Roja de la UICN), analizamos las causas de la declinaci´on y las clasificaciones de amenazas de la
¶Address correspondence to J. R. Mendelson III at Zoo Atlanta, email jmendelson@zooatlanta.org Paper submitted January 21, 2014; revised manuscript accepted May 13, 2015.
1
UICN al calcular las medidas filogen´eticas (Fritz y PurvisD). Tambi´en analizamos el alcance al cual la causa de la declinaci´on y la clasificaci´on de la amenaza se traslapan mediante el c´alculo la correlaci´on filogen´etica entre dos tipos de datos de caracter´ısticas. Las especies del g´eneroInciliusvariaron ampliamente tanto en la clasificaci´on de la amenaza como en la causa de la declinaci´on; variaci´on que se perder´ıa con una resoluci´on taxon´omica m´as burda. Encontramos muchas m´as se˜nales filogen´eticas, probablemente en correlaci´on con la historia evolutiva, para la causa de la declinaci´on que las clasificaciones de riesgo de la UICN. Las causas individuales de la declinaci´on y las clasificaciones de amenaza de la UICN no estuvieron correlacionadas en su mayor´ıa con la filogenia. Nuestros resultados demostraron la importancia de la selecci´on de car´acter y la resoluci´on taxon´omica cuando la amenaza de extinci´on es analizada en un contexto filogen´etico.
Palabras Clave:amenaza de extinci´on, anfibio, Bufonidae, filogenia, Mesoam´erica
Introduction
As extinction rates continue to increase (Pimm et al.
2014) and anthropogenic threats to biodiversity become better understood, a new consensus has emerged: the patterns and causes of decline tend not to be phylogenet- ically random (Purvis 2008). In particular, elevated levels of amphibian decline and extinction are now a well- corroborated phenomenon (Stuart et al. 2004; Wake &
Vredenburg 2008). The causes of amphibian declines are multiple, and surely are synergistic in many cases (Collins
& Storfer 2003; Sodhi et al. 2008; Collins & Crump 2009).
An assessment of the conservation status of all known amphibian species (Stuart et al. 2004) is available on the International Union for Conservation of Nature (IUCN) Red List (www.redlist.org), though the list is somewhat outdated in terms of threat assessment and taxonomic completeness. An important development since the pub- lication of this assessment is that a relatively complete and robust phylogeny of the major clades of amphibians now exists (Frost et al. 2006; Pyron & Wiens 2011).
These two classes of information—data regarding the level and types of threats faced by each species and a fully resolved, robust phylogeny with complete species- level sampling—are required for identification of phylo- genetic patterns in extinction risk. This has led to the use of amphibians in studies of phylogenetic patterns of extinction risk (e.g., Corey & Waite 2008; Bielby et al.
2008, 2010). These studies are based, at least in part, on the phylogeny of Frost et al. (2006) and information on species’ conservation status and threats from IUCN Red List. However, no complete phylogeny for the over 7000 species of amphibians exists. Phylogenetic studies of am- phibian extinction risk are therefore limited to higher tax- onomic categories, as is the case with other taxa (Brooks et al. 2005). Phylogenetic studies of threat ranking are therefore limited to broader patterns among major clades, which are often referred to as taxonomic families or gen- era in the literature (e.g., Bennett et al. 2005; Brooks et al. 2005). In such cases, the concept of taxonomic distinctiveness may be conflated with phylogenetic dis- tinctiveness (Avise 2005) inasmuch as supraspecific taxo- nomic categories are arbitrary and may be inconsistently
applied. Furthermore, clades of the same age can dif- fer greatly in species richness (Ricklefs et al. 2007); the amount of species-level data that is lost for each major clade (e.g., variation in IUCN threat rankings) may there- fore vary significantly between higher-level clades such as Linnaean families.
Another problem facing phylogenetic studies of threat ranking is the incompleteness or non-comparability of threat information available in the rank listings (e.g., IUCN Red List). The criteria and categorical definitions used by IUCN are intended to be as objective as possible (IUCN 2012), such that 2 species relegated to the same category (e.g., critically endangered) may be assumed to be in similar situations in terms of population trajectories and extent of geographic range occupied (either reduced by conservation threats or naturally very small). Between clades or geographic regions, comparisons may be made regarding the number of species in a category; however, it is not possible to reconcile the actual causes of de- cline with threat rankings in any standardized manner.
This means that if two species are ranked equivalently as critically endangered, one may be under threat from water pollution whereas the other may be threatened by over-harvesting for human consumption. In other words, although the equivalent ranks indicate that both species are similarly endangered, the causes can be wholly differ- ent (Davidson et al. 2009). In our opinion, this violates the concept of homology in phylogenetic systematics, in that one should not map critically endangered as a character onto a phylogeny because fundamentally differ- ent processes may underlie an equivalent threat ranking.
Homologous characters reflect similarity based on com- mon ancestry and are appropriate for use in the study of character evolution, whereas homoplastic characters are those that superficially appear to be similar but may have arisen through different mechanisms. For example, one would not code the character green coloration as equivalent in a frog and a bird because the equivalent condition is not produced by homologous mechanisms.
We approached the concept of phylogeny and conservation on a finer scale than most previous studies, which allowed us to use a taxonomically complete and fully resolved phylogeny of a clade, the Mesoamerican
toad genusIncilius(Mendelson et al. 2011). Our analysis of the genus begins with a replication of the methods used in previous studies: we calculated a number of autocorrelation metrics to measure phylogenetic signal in threat rankings. In addition to these conventional analyses, we also reconstructed the acknowledged cause or causes of decline to each species. Finally, we analyzed the correlation between different causes of decline, and between cause of decline and threat ranking. We present this as a case study demonstrating the advantages of working on a finer scale than most phylogeny- conservation studies and the advantages of considering cause of decline in the context of homology rather than using the threat rankings as a potentially homoplastic proxy.
Methods
We used three methods to determine whether phylogeny can be used to predict extinction risk. First, we calculated two traditional metrics of phylogenetic autocorrelation (Pagel’sλand Abouheif’s test), following the methods of previous studies of extinction risk in amphibians (Corey
& Waite 2008). Second, we calculated Fritz and Purvis’
(2010) D, a new metric that measures phylogenetic sig- nal of binary traits and may be better suited to discrete data. This metric has been used in more recent studies of phylogeny and extinction risk (Fritz & Purvis 2010;
Turvey & Fritz 2011; Yessoufou et al. 2012). Lastly, we measured the phylogenetic correlation of the different causes that underlie threat rankings and calculated the correlation between cause of decline and threat ranking.
Our analyses incorporated 39Inciliusspecies: all cur- rently recognized species exceptI. intermedius, whose taxonomic status is questionable; no natural population, extinct or extant, has been assigned to this taxon (Frost 2015). We used the phylogeny from Mendelson et al.
(2011) (their Fig. 6) as a starting point for our phylo- genies. Because several species ofInciliusevidently are extinct and several are extremely rare, some species were missing in the analyses presented by Mendelson et al.
(2011), so the authors inferred the phylogenetic positions of these species and illustrated them as minimally re- solved (as polytomies) in their final tree. We used that tree and further resolved those polytomies with the follow- ing justifications.Incilius mccoyiwas placed as sister to I. occidentalisbased on their similarity and evident rela- tionship (Santos-Barrera & Flores-Villela 2011). The clade ((guanacaste)(chompipe, epioticus)) was resolved to fol- low a hypothesized north-to-south vicariance model of speciation. Resolving it as a south-to-north model would have no effect on our analyses. Incilius gemmiferwas placed as sister toI. mazatlanensisbased on morpholog- ical similarity and a vicariant model of speciation.Incilius holdridgeiwas placed as sister toI. fastidiosusbased on
morphological and ecological similarity and proximate, but allopatric, distributions. The pairing of I. majordo- musandI. peripateteswas based on their close affiliation presented in the recent description of I. majordomus (Savage et al. 2013).
For branch lengths, we used the Bayesian analysis of the combined molecular and morphological data from Mendelson et al. (2011) (their Fig. 5). For the added taxa, we used branch lengths (brlen) comparable to their nearest neighbors. (I. mccoyibrlen=0.05 because I. occidentalis brlen = 0.0558; I. gemmifer brlen = 0.02 becauseI. mazatlanensis=0.0212;I guanacaste 0.04 because “Crepdiophryne” was 0.0475. ForI. chom- pipeandI. epioticus, we used brlen =0.0475 because the “Crepdiophryne” of Mendelson et al. [2011] was a chimera ofepioticusandchompipe.I. periglenes, I. ma- jordomus, I. peripatetes,andI. holdridgeiwere all set to brlen=0.04 becauseI. fastidiosus=0.0467.) We used the value 0.001 to connect branch lengths of all added taxa. To make the tree ultrametric, we used TreeEdit (Rambaut & Charleston 2001) and chose the nonpara- metric rate smoothing method (Sanderson 1997). Be- cause of uncertainty in the immediate outgroup to Incilius(Mendelson et al. 2011), we conducted all anal- yses with 2 trees: the first, henceforth RR, rooted on the branch leading to theRhaebo–Rhinella clade, and the second, henceforth Ana, rooted on the branch leading to Anaxyrus(Fig. 1). All tests of phylogenetic signal were performed on both phylogenies.
We used two sets of threat rankings: IUCN (2014) Red List and a new data set (hereafter updated rankings) con- taining our re-assessment of the conservation status of the included species (but using IUCN categories) and our evaluation of the general consensus of the principal cause or causes of decline for each species (Supporting Infor- mation). The new data set was compiled based on our own field experience with these toads; nevertheless, our conservation-threat evaluations were largely consistent with the IUCN Red List. For each data set, we compiled binary data matrices for each threat ranking: not evalu- ated/data deficient, least concern, near threatened, vul- nerable, endangered, critically endangered, and extinct.
The threat rankings endangered, critically endangered, and extinct were also analyzed jointly in a severe threat category, as has been done in previous studies (Corey &
Waite 2008). In addition, we compiled binary data matri- ces for the three categories of principal cause of decline inInciliusspecies: small range, habitat destruction, and a pathogenic chytrid fungus (Batrachochytrium dendro- batidis). For principal cause of decline, unlike the IUCN and updated threat rankings, a single species can belong to more than one category. We included small range as a cause of decline because the correlation between small range and extinction probability is a well-supported and generally accepted phenomenon (Purvis et al. 2000). If a species has very small range then its entire habitat can
I.nebulifer I.gemmifer*
I.valliceps I.mazatlanensis
I.luetkenii
I.nebulifer I.gemmifer*
I.valliceps I.mazatlanensis
I.luetkenii I.aucoinae
I.guanacaste*
I.marmoreus I.epioticus I.melanochlorus
I.bocourti I.cycladen I.campbelli
I.peripatetes* I.pisinnus I.porteri
I.ibarrai I.cavifrons
I.periglenes* I.holdridgei*
I.perplexusI.perplexus I.fastidiosus I.cristatus
I.coniferus I.coccifer
I.tacanensis I.canaliferus I.karenlipsae
I.majordomus*
I.occidentalis I.mccoyi*
I.alvarius I.signifer I.tutelarius I.aurarius
I.leucomyos
I.chompipe
I.guanacaste* I.epioticus
I.peripatetes*
I.periglenes* I.holdridgei*I.fastidiosus I.coniferus I.karenlipsae
I.majordomus* I.chompipe I.macrocristatus
I.spiculatus
coccifer group coniferus
group
vallicepsgroup
vallicepsgroup Forest toads
I.tutelarius
I.bocourti I.signifer I.porteri
I.cycladen I.leucomyos I.macrocristatus
I.mccoyi* I.canaliferus I.pisinnus
I.tacanensis I.coccifer I.campbelli
I.ibarrai
I.marmoreus I.aucoinae I.cavifronsI.cristatusI.melanochlorus
I.alvarius I.occidentalis I.spiculatus I.aurarius
coccifer group coniferus
group Forest toads
Outgroup: Rhinella, Rhaebo
Outgroup: Anaxyrus
R.haematiticus R.schneideri
R.festae R.margaritifera A.boreas
A.boreas R.marina
R.haematiticus R.festae R.margaritifera R.schneideriR.marina
Incilius Incilius
Figure 1. The 2Inciliusphylogenies, with different outgroups, analyzed (adapted from Mendelson et al.’ [2011]
Fig. 6) with resolved polytomies. The dashed lines and asterisks indicate post-hoc placement of species for which no samples of DNA are known to exist and that were therefore not included in the original analyses by Mendelson et al. (2011) (see Methods for justification).
very easily be destroyed, and we considered this a threat to the species. In the case ofIncilius, several species are known only from their type locality.
Phylogenetic Signal
We used 2 traditional tests of phylogenetic autocor- relation: Pagel’s λ (Pagel 1999) and Abouheif’s test (Abouheif 1999). Pagel’s λ detects random and Brown- ian phylogenetic distributions of continuous traits well (M¨unkem¨uller et al. 2012) and is one of the few mea- sures of phylogenetic signal that can be used to measure discrete traits. Abouheif’s test is a qualitative implemen- tation of Moran’sI(Pavoine et al. 2008), which has been used in previous studies of extinction risk in amphib- ians (Corey & Waite 2008) and has been recommended for such studies (Hardy et al. 2012). We calculated both metrics inR(R Development Core Team 2013) with the geiger package for Pagel’sλ(Harmon et al. 2008) and theadephylopackage for Abouheif’s test (Jombart et al.
2010). We then used a likelihood ratio test to compare the negative log likelihoods of two values ofλ: the maximum likelihood estimate ofλand aλof zero (which removes branch lengths, transforming the tree into a polytomy and thus erasing all phylogenetic signal).
Another, more recent measure of phylogenetic signal in extinction threat is Fritz and Purvis’D(Fritz & Purvis 2010), which we also calculated in R with the caper package (Orme 2012; R Development Core Team 2013).
The resultingpvalue measures whetherDis significantly
<1, indicating strong phylogenetic signal.
Because a single species can fall into multiple cat- egories of causes of decline, we calculated the phy- logenetic correlation of different causes of decline in theDiscretemodule of BayesTraits(Pagel & Meade 2013). Using the same methods, we also performed correlation analysis between each threat ranking and each cause of decline. We compared two models: a 4-parameter model of independent evolution and an 8-parameter model of dependent evolution. Analyses were performed with a maximum likelihood algorithm.
We used a likelihood ratio test approximated by a chi- squared distribution with four degrees of freedom and compared the negative log likelihoods of the indepen- dent and dependent models of evolution.
For all the metrics calculated here, we used binary data matrices to examine 1 or 2 characters in isolation;
therefore, the total number of characters considered did not affect our results. For example, the inclusion of small range as a cause of decline had no effect on the p values obtained for habitat destruction and B. den- drobatidis. Similarly, the inclusion of the severe threat category (combined endangered, critically endangered, and extinct) had no effect on thep values obtained for the other threat rankings.
Table 1. Phylogenetic autocorrelation, based on Pagel’s λ and Abouheif’s tests, of extinction risk and cause of decline inIncilius toads and 2 selected outgroups. Calculated two traditional metrics of phylogenetic autocorrelation.
Pagel’sλa Abouheif’s testa
RR Ana RR Ana
Cause of decline
habitat destruction 0.000∗∗∗ 0.000∗∗∗ 0.001∗∗ 0.001∗∗
small range 0.033 0.034 0.001∗∗ 0.001∗∗
Chytridiomycosis 0.895 0.505 0.019 0.006∗ IUCN rankb
NE, DD 1.000 1.000 0.793 0.781
LC 0.039 0.038 0.108 0.102
NT 0.532 0.321 0.146 0.069
VU 1.000 1.000 0.912 0.894
EN 1.000 1.000 0.168 0.169
CR 1.000 1.000 0.011 0.015
EX 1.000 1.000 0.532 0.553
EN-CR-EX 1.000 1.000 0.047 0.053
Updated rankb
NE, DD 1.000 1.000 0.834 0.815
LC 0.024 0.023 0.011 0.010
NT 0.116 0.073 0.012 0.005∗
VU 1.000 1.000 0.039 0.040
EN 1.000 1.000 0.892 0.896
CR 1.000 1.000 0.055 0.049
EX 1.000 1.000 0.175 0.186
EN-CR-EX 1.000 1.000 0.018 0.016
aPhylogenies: RR,Inciliusphylogeny rooted withRhaeboandRhinella as outgroups; Ana,Inciliusphylogeny rooted withAnaxyrusas the outgroup. Significance:∗p<0.05;∗∗p<0.005;∗∗∗p<0.001.
bThe IUCN Red List categories: NE, not evaluated; DD, data deficient; LC, least concern; NT, near threatened; VU, vulner- able; EN, endangered; CR, critically endangered; EX, extinct.
None of the taxa analyzed here are extinct in the wild (EW).
Results
We obtained fairly similar results from the 2 differently rooted phylogenies and from both sets of threat rank- ings. With Pagel’sλ(Table 1), we found four significantly autocorrelated traits for each tree: habitat destruction as a cause of decline, small range as a cause of decline, least concern threat ranking according to the IUCN, and least concern threat ranking according to our new data set. However, the threat-ranking traits were no longer sig- nificantly autocorrelated when the Bonferroni correction was applied. Habitat destruction as a cause of decline was a very highly autocorrelated trait on both phylogenies (p<0.0001). This was the only trait that remained sig- nificantly autocorrelated when the Bonferroni correction was applied to the data.
Abouheif’s test was the only measure for which we found significant differences between the two phylogenies (Table 1). This test was also far less conservative than Pagel’s λ because it showed the highest number of significantly autocorrelated traits.
However, when the Bonferroni correction was applied,
Table 2. Fritz and Purvis’Das a measure of phylogenetic signal in extinction risk and cause of decline ofInciliustoads.
RRa Anaa
D p D p
Cause of decline habitat
destruction
−0.986 <0.001∗∗∗ −1.068 <0.001∗∗∗
small range −2.548 <0.001∗∗∗ −2.613 <0.001∗∗∗
Chytridiomycosis −0.062 0.018 −0.154 0.018 IUCN rankb
NE, DD 1.507 0.895 1.495 0.869
LC 0.628 0.130 0.618 0.129
NT 0.397 0.181 0.379 0.205
VU 1.701 0.857 1.669 0.836
EN 0.743 0.265 0.766 0.264
CR −0.313 0.033 −0.348 0.031 EX −2.401 0.085 −1.295 0.077
EN-CR-EX 0.464 0.070 0.438 0.072
Updated rankb
NE, DD 4.815 0.869 3.106 0.872
LC 0.209 0.014 0.196 0.014
NT −0.587 0.009 −0.776 0.008
VU 0.195 0.057 0.214 0.054
EN 1.017 0.475 1.002 0.467
CR 0.240 0.045 0.224 0.046
EX 0.575 0.201 0.559 0.221
EN-CR-EX 0.221 0.017 0.202 0.019
aPhylogenies: RR,Inciliusphylogeny rooted withRhaeboandRhinella as outgroups; Ana,Inciliusphylogeny rooted withAnaxyrusas the outgroup. The p values indicate the probability that D<1 (departure from random phylogenetic distribution of the trait).
bThreat categories are identified in Table 1.
only one threat ranking (near threatened under the updated rankings) was significantly autocorrelated.
Habitat destruction and small range yielded significant pvalues with Pagel’sλand Abouheif’s test, and negative values ofD(Table 2). These are the only two traits for which the probability of a Brownian phylogenetic distri- bution was>0.95 under Fritz and Purvis’Dand are the only two traits that remained significantly autocorrelated when the Bonferroni correction was applied to the re- sults ofD. Chytrid fungus as a cause of decline yielded apof 0.018 with both trees; this value was marginally non-significant with the Bonferroni correction. Threat rankings that yield significant p values resulting from Fritz and Purvis’D—critically endangered according to the IUCN and a number of threat rankings under our new data set (Table 1)—were no longer significant when the Bonferroni correction was applied.
With or without the Bonferroni correction, habitat de- struction and small range were the only cause-of-decline categories that were significantly correlated with each other (Supporting Information). Among cause of decline and threat ranking correlations, 12 of the 96 pairings yielded significant p values without the Bonferroni correction; however, with the Bonferroni correction, none of these results were significant (Table 3). When the
Table 3. Phylogenetic correlation between cause of decline and threat level inInciliustoads.
RRa Anaa
Bd HD SR Bd HD SR
IUCN rankb
NE, DD 0.568 0.582 0.809 0.528 0.561 0.848 LC 0.061 0.193 0.791 0.058 0.199 0.837 NT 0.688 0.291 0.787 0.287 0.343 0.949 VU 0.904 0.798 0.623 0.907 0.805 0.663 EN 0.575 0.837 0.794 0.575 0.837 0.834 CR 0.032 0.875 0.364 0.033 0.861 0.340 EX 0.377 0.769 0.949 0.380 0.802 0.979 EN-CR-EX 0.028 0.861 0.620 0.027 0.855 0.648 Updated rank
NE, DD 0.481 0.769 0.956 0.411 0.802 0.979 LC 0.070 0.001 0.442 0.081 0.002 0.472 NT 0.620 0.333 0.894 0.487 0.333 0.949 VU 0.565 0.009 0.300 0.681 0.009 0.324 EN 0.954 0.289 0.475 0.956 0.275 0.506 CR 0.264 0.512 0.812 0.269 0.451 0.855 EX 0.009 0.258 0.264 0.010 0.275 0.254 EN-CR-EX 0.002 0.409 0.666 0.002 0.377 0.670
aPhylogenies: RR, the Incilius phylogeny rooted with Rhaebo and Rhinella as outgroups; Ana, the Incilius phylogeny rooted with Anaxyrusas the outgroup. Other abbreviations: Bd:B. dendrobatidis, commonly known as the amphibian chytrid fungus; HD, habitat destruction; SR, small range.
bThreat categories are identified in Table 1.
Bonferroni correction was not applied, the significant results were the same from both phylogenies. Notably, chytrid fungus was associated with severe threat (endangered, critically endangered, or extinct), and small range, a rare cause of decline (affecting only 3 Incilius species), was not significantly associated with any single threat ranking.
Discussion
We found far more phylogenetic signal in cause of decline than in threat ranking. Given that 96 tests of phyloge- netic autocorrelation were conducted on threat ranking (Pagel’sλ, Abouheif’s test, and Fritz and Purvis’D; two phylogenetic hypotheses; two data sets of rankings; eight rankings), one would expect between 4 and 5 false pos- itives atα =0.05. Yet, with the Bonferroni correction, only one of these 96 tests yielded a significantp value.
Our results therefore strongly suggest that there is no species-level phylogenetic signal in threat ranking within Incilius; the different threat rankings were widely dis- persed across the phylogeny (Supporting Information).
In contrast, a very different picture emerges from the results pertaining to cause of decline (Fig. 2). All metrics very strongly supported the designation of an Incilius clade sensitive to habitat destruction. Two of our three metrics we used strongly supported the designation of
Outgroup: Rhinella, Rhaebo
Outgroup: Anaxyrus
I.aucoinae
I.guanacaste I.epioticus
I.peripatetes
I.periglenes I.holdridgeiI.fastidiosus I.coniferus I.karenlipsae
I.majordomus I.chompipe
I.guanacaste I.epioticus
I.peripatetes
I.periglenes I.holdridgeiI.fastidiosus I.coniferus I.karenlipsae
I.majordomus I.chompipe
I.marmoreus I.melanochlorus
I.bocourti
R.haematiticus R.schneideri I.cycladen I.campbelli
R.festae I.nebulifer I.gemmifer
I.valliceps I.mazatlanensis
I.luetkenii
I.nebulifer I.gemmifer
I.valliceps I.mazatlanensis
I.luetkenii
I.pisinnus I.porteri
I.ibarrai I.cavifrons
I.perplexusI.perplexus I.cristatus
R.margaritifera I.coccifer
I.tacanensis
A.boreas
R.marina I.canaliferus
I.occidentalis I.mccoyi
I.alvarius I.signifer I.tutelarius I.aurarius
I.leucomyos I.macrocristatus I.spiculatus I.tutelarius
A.boreas I.bocourti
R.haematiticus I.signifer
R.festae I.porteri
I.cycladen
R.margaritifera I.leucomyos I.macrocristatus
I.mccoyi
R.schneideri I.canaliferus I.pisinnus
I.tacanensis I.coccifer I.campbelli
I.ibarrai
I.marmoreus I.aucoinae
R.marina I.cavifronsI.cristatusI.melanochlorus
I.alvarius I.occidentalis I.spiculatus I.aurarius
Amphibian chytrid fungus Habitat destruction
Small range and habitat destruction Unknown
None
Figure 2. Causes of decline inInciliusin a phylogenetic context.
a small-range-endangered Incilius clade. The 3 species that are threatened due to small range —I. guanacaste, I. chompipe, and I. epioticus—formed a monophyletic clade in both phylogenies; we therefore suspect that the lack of strong support from Pagel’s λ was due to the small size of this clade. Our correlation analysis showed a high degree of overlap between these 2 causes of decline (habitat destruction and small range), which is unsurprising because the 3 species threatened due to small range are also threatened due to habitat destruc- tion. The clade that is sensitive to small range and habi- tat destruction was not nested within the clade that is sensitive to habitat destruction only. Our results support the designation of anIncilius clade that is endangered due to habitat destruction—a group often referred to as the forest toads—with a separate clade nested within the coniferusgroup that is endangered due to both habitat destruction and small range. It is highly unlikely that this phylogenetic signal is an artifact of spatial autocorrelation because mostInciliusspecies are parapatric or allopatric and closely related species are particularly unlikely to co-occur in a given area (Supporting Information). We are aware of only 2 instances of syntopy (I. valliceps andI. marmoreus, in Veracruz, Mexico, andI. perplexus and I. marmoreus in west-central Mexico). However, researchers using our method would want to take spatial autocorrelation into account if they are studying species that show significant amounts of sympatry.
The data regarding the chytrid fungus were incon- clusive because different tests yielded non-significant, marginally significant, and significant p values for au- tocorrelation of this trait. We cannot confidently say whether or not the data supported the designation of a chytrid-endangered clade within Incilius.Smith et al.
(2009) did find such a pattern working at the community level in Costa Rica and Panama. InIncilius, the chytrid fungus threatens I. ibarrai, which is in the coccifer group, as well as a number of species within theconiferus group. Together, thecocciferandconiferusgroups form a monophyletic clade. The notable lack of support for a chytrid-endangered clade according to the Pagel’sλmay be due to the fact that one outgroup lineage,A. boreas, is also endangered due to the chytrid fungus.
In addition to our autocorrelation results, which show that cause of decline and threat ranking contain different amounts of phylogenetic signal, the results of our corre- lation analysis strongly suggest that cause of decline is decoupled from threat ranking. This is because cause of decline and threat ranking tend not to be correlated, such that threat ranking (which is a homoplastic character) ap- pears not to be an appropriate proxy for cause of decline (a homologous character).
A recurring problem for estimates of phylogenetic sig- nal in conservation-threat assessment is the lack of a com- plete and fully resolved phylogeny. Some researchers had access to nearly complete phylogenies (e.g., Isaac et al.
2007, Mammalia) or created a complete phylogeny of focal taxa and documented well or experimentally con- trolled causes of threat (e.g., Smith et al. 2009; Hammond et al. 2012). Those studying other groups, such as the am- phibians, had to seek patterns at a broad scale because of incomplete sampling or the coarse resolution of the phy- logeny available to them. Because we found that signifi- cant differences in threat ranking existed at the species level, our results demonstrate the importance of com- plete sampling and fine resolution when searching for phylogenetic signal in threat rankings. Results at broader scales tend to be relatively uninformative. For example, a key result from Corey and Waite (2008) is that certain fam- ilies of hyloid frogs are statistically correlated in terms of the IUCN threat ranking critically endangered and cause of decline is “enigmatic decline” (latter data from Stuart et al. 2004). Thus, taxonomic families such as Bufonidae (approximately 575 species, worldwide distribution) are flagged wholesale as critically endangered by enigmatic declines as a result of their shared evolutionary history.
This conclusion clearly is inapplicable to many species within Bufonidae. At a smaller phylogenetic scale, with better-defined causes of declines, we identified phyloge- netic patterns that should inform protection and recovery programs far better. For example, our results clearly indi- cate a clade in need of progressive habitat conservation programs and suggest that another clade may bear the synapomorphy, or shared trait inherited from a common ancestor, of special sensitivity to chytrid fungal infec- tions. Our data also highlight interesting exceptions that may be important in terms of conservation. For example, the speciesI. gemmifer, from Mexico, has not been found in over 40 years, despite repeated searches, yet its threat ranking as either critically endangered or extinct would not be predicted by its phylogenetic position.
We also found that homologous characters pertaining to cause of extinction contained strong phylogenetic signal, whereas homoplastic data pertaining to severity of extinction threat (e.g., IUCN threat rankings) did not reflect phylogeny. This difference between homologous and homoplastic characters has implications for the appli- cation of evolutionary history to conservation. Although our study is limited to a single genus, our findings are highly relevant to a large number of studies that report potential effects of shared evolutionary history in analyses of susceptibility or that identify clade-level influences of specific conservation threats (e.g., Cooper et al. 2007;
Bielby et al. 2008; Davies et al. 2008; Isaac et al. 2012).
Our approach could be applied to any taxon, and ad- ditional studies should examine the characters consid- ered here among species in larger clades. We suggest that future studies incorporate the important distinction between the homoplastic trait “proneness to equivalent IUCN rankings” as opposed to the homologous trait
“proneness to extinction due to shared susceptibility to specific causes of decline.”
The IUCN threat rankings are essential for purposes of immediate, short-term conservation activities. However, phylogenies reflect evolutionary history over very long periods and are based on the concept of homology. Our results strongly suggest that phylogenetic studies meant to guide long-term conservation strategy should use ho- mologous characters relating to cause of decline, rather than homoplastic designations derived from IUCN threat rankings.
Acknowledgments
We thank J.C. Maerz, A. Belasen, C. VanBuren, and two anonymous reviewers for constructive feedback and dis- cussions. S.R.S. is supported by the National Science Foundation Graduate Research Fellowship Program un- der grant DGE-1125191.
Supporting Information
The current IUCN Red List categories for Inciliusspp., our recommended updated categories for those species, and principal cause of decline (Appendix S1); phy- logenetic correlation of cause of decline of Inclius toads (Appendix S2); rankings of extinction risk for Incilius (Appendix S3); and general distributions ofIn- cilius species (Appendix S4) are available online. The authors are solely responsible for the content and func- tionality of these materials. Queries (other than absence of the material) should be directed to the corresponding author.
Literature Cited
Abouheif E. 1999. A method for testing the assumption of phylogenetic independence in comparative data. Evolutionary Ecology Research 1:895–909.
Avise JC. 2005. Phylogenetic units and currencies above and below the species level. Pages 76–100 in Purvis A, Gittleman JL, Brooks T, editors. Phylogeny and conservation. Cambridge University Press, New York.
Bennett PR, Owen IPF, Nussey D, Garnett ST, Crowley GM. 2005. Mech- anisms of extinction in birds: phylogeny, ecology, and threats. Pages 317–336 in Purvis A, Gittleman JL, Brooks T, editors. Phylogeny and conservation. Cambridge University Press, New York.
Bielby J, Cooper N, Cunningham AA, Garner TWJ, Purvis A. 2008.
Predicting susceptibility to future declines in the world’s frogs.
Conservation Letters1:82–90.
Bielby J, Cardillo M, Cooper N, Purvis A. 2010. Modelling extinction risk in multispecies data sets: phylogenetically independent con- trasts versus decision trees. Biodiversity and Conservation19:113–
127.
Brooks TM, Pilgrim JD, Rodrigues ASL, Fonseca GABD. 2005. Conser- vation status and geographic distribution of avian evoloutionary history. Pages 267–294 in Purvis A, Gittleman JL, Brooks T, edi- tors. Phylogeny and conservation. Cambridge University Press, New York.
Collins JP, Storfer A. 2003. Global amphibian declines: sorting the hy- potheses. Diversity and Distribution9:89–98.
Collins JP, Crump ML. 2009. Extinction in our times: global amphibian decline. Oxford University Press, New York.
Cooper N, Bielby J, Thomas GH, Purvis A. 2007. Macroecology and extinction risk of frogs. Global Ecology and Biogeography17:211–
221.
Corey SJ, Waite TA. 2008. Phylogenetic autocorrelation of extinction threat in globally imperiled amphibians. Diversity and Distributions 14:614–629.
Davidson AD, Hamilton MJ, Boyer AG, Brown JH, Ceballos G. 2009. Mul- tiple ecological pathways to extinction in mammals. Proceedings of the National Academy of Science (USA)106:10702–10705.
Davies TJ, et al. 2008. Phylogenetic trees and the future of mammalian biodiversity. Proceedings of the National Academy of Science (USA) 105:11556–11563.
Fritz SA, Purvis A. 2010. Selectivity in mammalian extinction risk and threat types: a new measure of phylogenetic signal strength in binary traits. Conservation Biology24:1042–1051.
Frost DR. 2015. Amphibian species of the world: an online refer- ence. Version 6.0. American Museum of Natural History, New York. Available from http://research.amnh.org/vz/herpetology/
amphibia/index.php (accessed April 2015).
Frost DR, et al. 2006. The amphibian tree of life. Bulletin of the American Museum of Natural History297:1–370.
Hammond JI, Jones DK, Stephens PR, Relyea RA. 2012. Phy- logeny meets ecotoxicology: evolutionary patterns of sensitivity to a common insecticide. Evolutionary Applications 5:593–606.
DOI:10.1111/j.1752-4571.2011.00237.x.
Hardy C, Fara E, Laffont R, Dommergues J-L, Meister C, Neige P. 2012.
Deep-time phylogenetic clustering of extinctions in an evolutionar- ily dynamic clade (Early Jurassic ammonites). PLOS ONE7:e37977 DOI: 10.1371/journal.pone.0037977.
Harmon LJ, Weir JT, Brock CD, Glor RE, Challenger W. 2008. GEIGER:
investigating evolutionary radiations. Bioinformatics24:129–131.
Isaac NJB, Tuvey ST, Collen B, Waterman C, Bailie JEM. 2007. Mammals on the EDGE: conservation priorities based on threat and phylogeny.
PLOS ONE2: e296 DOI:10.1371/journal.pone.0000296.
Isaac NJB, Redding DW, Meredith HM, Safi K. 2012. Phylogentically- informed priorities for amphibian conservation. PLOS ONE 7:
e43912 DOI:10.1371/journal.pone.0043912.
IUCN (International Union for Conservation of Nature). 2012. IUCN red list categories and criteria. Version 3.1. 2nd edition. IUCN, Gland.
IUCN (International Union for Conservation of Nature). 2014. The IUCN red list of threatened species. Gland, Switzerland. Available from http://www.iucnredlist.org/about/citing (accessed September 2014).
Jombart T, Balloux F, Dray S. 2010. adephylo: new tools for investigating the phylogenetic signal in biological traits. Bioinformatics26:1907–
1909.
Mendelson III JR, Mulcahy DG, Williams TS, Sites Jr JW. 2011. A phylogeny and evolutionary natural history of Mesoamerican toads (Anura: Bufonidae: Incilius) based on morphology, life history, and molecular data. Zootaxa3138:1–34.
M¨unkem¨uller T, Lavergne S, Bzeznik B, Dray S, Jombart T, Schiffers K, Thuiller W. 2012. How to measure and test phylogenetic signal.
Methods in Ecology and Evolution3:743–756.
Orme C. 2012. The caper package: comparative analyses in phylogenet- ics and evolution in R. Vienna, Austria. Available from http://caper.r- forge.r-project.org/ (accessed September 2014).
Pagel M. 1999. Inferring the historical patterns of biological evolution.
Nature401:877–84.
Pagel M, Meade A. 2013. BayesTraits. Reading, United Kingdom. Avail- able from www.evolution.rdg.ac.uk (accessed September 2014).
Pavoine S, Ollier S, Pontier D, Chessel D. 2008. Testing for phylogenetic signal in phenotypic traits: new matrices of phylogenetic proximi- ties. Theoretical Population Biology73:79–91.
Pimm SL, Jenkins CN, Abell R, Brooks TM, Gittleman JL, Joppa IN, Raven PH, Roberts CM, Sexton JO. 2014. The biodiversity of species and their extinction, distribution, and protection. Science344:987–
996.
Purvis A. 2008. Phylogenetic approaches to the study of extinction.
Annual Review of Ecology and Systematics39:301–319.
Purvis A, Gittleman JL, Cowlishaw G, Mace GM. 2000. Predicting ex- tinction risk in declining species. Proceedings of the Royal Society B267:1947–1952.
Pyron RA, Wiens JJ. 2011. A large-scale phylogeny of Amphibia includ- ing over 2800 species, and a revised classification of extant frogs, salamanders, and caecilians. Molecular Phylogenetics and Evolution 61:543–583.
Rambaut A, Charleston M. 2001. TreeEdit: phylogenetic tree editor. Oxford, United Kingdom. Available from http://tree.
bio.ed.ac.uk/software/treeedit/main.
R Development Core Team. 2013. R: A language and environment for statistical computing. Vienna, Austria. Available from www.r- project.org (accessed September 2014).
Ricklefs RE, Losos JB, Townsend TM. 2007. Evolutionary diversifica- tion of clades of squamate reptiles. Journal of Evolutionary Biology 20:1751–1762.
Sanderson MJ. 1997. A nonparametric approach to estimating diver- gence times in the absence of rate constancy. Molecular Biology and Evolution14:1218–1231.
Santos-Barrera G, Flores-Villela O. 2011. A new species of toad of the genusInciliusfrom the Sierra Madre Occidental of Chihuahua, Mex- ico (Anura: Bufonidae). Journal of Herpetology45:211–215.
Savage JM, Ugarte CA, Donnelly MA. 2013. A new species of earless toad (Bufonidae:Incilius) from western Panama. Copeia2013:8–12.
Smith KG, Lips KR, Chase JM. 2009. Selecting for extinction: non- random disease-associated extinction homogenizes amphibian bio- tas. Ecology Letters12:1069–1078.
Sodhi NS, Bickford D, Diesmos AC, Lee TM, Koh LP, Brook BW, Seker- cioglu CH, Bradshaw CJA. 2008. Measuring the meltdown: drivers of global amphibian extinction and decline. PLOS ONE 3:e1636 DOI:10.1371/journal.pone.0001636.
Stuart SN, Chanson JS, Cox NA, Young BE, Rodrigues ASL, Fischman DL, Waller RW. 2004. Status and trends of amphibian declines and extinctions worldwide. Science306:1783–1786.
Turvey ST, Fritz SA. 2011. The ghosts of mammals past: biological and geographical patterns of global mammalian extinction across the Holocene. Philosophical transactions of the Royal Society of London. Series B, Biological Sciences366:2564–76.
Wake DB, Vredenburg VT. 2008. Are we in the midst of the sixth mass extinction? A view from the world of amphibians. Proceedings of the National Academy of Science (USA)105:11466–11473.
Yessoufou K, Daru BH, Davies TJ. 2012. Phylogenetic patterns of ex- tinction risk in the eastern arc ecosystems, an African biodiversity hotspot. PLOS ONE7:e47082 DOI:10.1371/journal.pone.0047082.
